By Zoe Karkossa with Louis Doré-Savard
Integrative neuroscience is a flourishing approach to understanding the human brain that carefully weaves together a wide variety of different data and analytical methods. Neurological research has traditionally adopted an atomic approach that selectively considers a single protein, gene, transcript, neurotransmitter or neuroimaging modality at a time. However, there is a great amount of information to be gained by studying the complex interactions that govern these innumerable factors across multiple scales.
The Neuroinformatics for Personalized Medicine Lab at the McGill Neurological Institute and Hospital (The Neuro) studies how computational integrational methods can be used to understand neuroscience and neurodegeneration at the individual level. Led by Yasser Iturria-Medina, PhD, the team works on the development of a variety of intricate models that allow insight into the complex functioning of each single brain. This neuroinformatic approach is a unique application of mathematical and engineering concepts to molecular and neuroimaging data.
“Normally in neurology, the studies that dominate are usually population studies. When you look at the data behind most publications, you see a lot of variability [between subjects].” Iturria-Medina said in an interview with the Ludmer Centre. “That variability is extremely important, because it determines actually the causes of the individual disease and who will [respond] or not to a given treatment.”
Alzheimer's Disease and Dementia in Canada
Over 747,000 Canadians are living with Alzheimer's or another dementia. Worldwide, at least 44 million people are living with dementia—more than the total population of Canada—making the disease a global health crisis that must be addressed.
In two upcoming papers, Yasser and his team explore both a highly integrative modeling approach and open-access software that can bring similar models to clinicians and other researchers. The focus is on the presentation, progression and potential treatment of Alzeimer’s disease (AD), the most common form of dementia. Despite decades of research, there is still no effective form of prevention or cure for the illness, in part due to its great complexity.
AD lends itself well to integrative modeling since it is characterized by a wide range of disturbed neurological processes across different scales, from the accumulation of amyloid plaques to vascular abnormalities. Neuroinformatic methods can allow insight into disease progression on an individual level and could potentially help identify favorable treatment plans. Identifying and exploring which therapies can work for a specific person, although they might not be a perfect remedy on a population level, is a critical step to improving both longevity and quality of life for patients.
“If we are able to develop comprehensive mathematical models that go from the molecular to the macroscopic [entire brain] scale and reflect how changes in one scale affects changes on the other biological scale, that could be very helpful [in] develop[ing] therapies,” Iturria-Medina said. “You could, for example, create computational models to predict in advance what would happen with a specific gene therapy.”
Dr. Yasser Iturria-Medina
Dr. Iturria-Medina is an Assistant Professor in the Department of Neurology and Neurosurgery. He is also an associate member of the Ludmer Centre for Neuroinformatics and Mental Health and the McConnell Brain Imaging Centre. His lab pursues primarily the goal of making precision medicine in Neurology a reality. It focuses on defining and implementing multiscale and multifactorial brain models for understanding neurological disorders and identifying effective personalized interventions.
Integrated Transcriptomic and Neuroimaging Brain Model Decodes Biological Mechanisms in Aging and Alzheimer’s Disease, first author PhD candidate Quadri Adewale, is currently online with eLife. The paper presents an innovative spatiotemporal model that combines a wide variety of data sources to provide insight into the functioning of both healthy aging and Alzheimer’s disease-affected brains.
The Gene Expression Multifactorial Causal Model (GE-MCM) integrates transcriptomic data, which consists of the hundreds of RNA transcripts present in the brain and their interactions, with positron emission tomography (PET) and magnetic resonance imaging (MRI) neuroimaging data. Disease progression is modeled through the combination of both local and network-level interactions. Although the data from many individuals can be combined, the key advance is the ability to incorporate many different modalities and scales for a single brain.
“If we integrate [all] the possible scales of data for each person, there is an immensely rich amount of information that needs to be analyzed in a unifying way,” Iturria-Medina said.
The results confirm and extend previous work done elucidating genes related to both aging and AD, as well as identifying new ones. Using data from almost 1000 individuals, GE-MCM successfully identifies many key genes related to neurodegenerative processes. Included are the key genes that determine the accumulation of amyloid-beta and tau, proteins that interact synergistically in speeding up neural loss, as well as the cerebral blood flow that enhances misfolded proteins clearance.
The model was also used to compare the state of biological pathways in both healthy aging individuals and those with AD. Although many of these mechanisms were shared, GE-MCM effectively recognized the more advanced alteration of many genetic mechanisms in the patient population. Overall, it provides many novel insights into both neurodegenerations causing AD as well as the state of healthy aging brains.
The second paper, to be published in Nature’s Communications Biology on May 21st, presents a used-friendly open-access software integrating various models and data sources that Iturria-Medina and his team have developed. Integrating Molecular, Histopathological, Neuroimaging and Clinical Neuroscience data with NeuroPM-box highlights the variety of data sources that can be flexibly incorporated using the toolbox.
The Neuroinformatics for Personalized Medicine toolbox (NeuroPM-box) allows the combination of data across molecular, histopathological brain-imaging and clinical evaluation modalities to model brain behavior and activity. It is designed to allow a unifying approach to brain analysis that is accessible by clinicians and researchers alike. The many models included were all extensively validated with previous work in the field and optimized for the application. A long-term work still in progress, Iturria-Medina hopes to continue finetuning the software and adding different modalities and scales given user feedback.
“One strong motivation [for open-access] is to give others the tools that we have developed, if not they will not be used [to their full capacity]. [Our lab] cannot cover all the applications and all the potential that could be done.” Iturria-Medina said.
Both papers were published in open access journals. eLife is a high impact journal that will hopefully allow the models to reach a wide variety of people in the field. This allows more clinicians researchers to use the methods to advance their own work, but also to explore new avenues and improve the models with feedback. As state-of-the-art research, there is still great room for improvement and expansion.
“The models are being extended to incorporate more types of different data, including epigenetics, neurotransmitter receptors, and metabolomics”, Iturria-Medina said. “In parallel to that, we need to also be sure that everything that we add is solid. We are validating and re-validating these models that are original and previous models with clinical trial data […] to check if they really predict brain responses, and if they can really clarify different disorders and not only Alzheimer’s.”
Both papers represent important advances in the movement towards an integrational view of neuroscience in an open science context. Iturria-Medina hopes to continue expanding work in the field to further research and improve effective treatment of neurodegenerative diseases. NeuroPM-box is available here.
To learn more:
Integrated transcriptomic and neuroimaging brain model decodes biological mechanisms in aging and Alzheimer’s disease. eLife. 2021 May 18;10:e62589. doi: 10.7554/eLife.62589
Integrating Molecular, Histopathological, Neuroimaging and Clinical Neuroscience data with NeuroPM-box. Y. Iturria-Medina, F. Carbonell, A. Assadi, Q. Adewale, A.F. Khan, R. Baumeister and L. Sanchez-Rodriguez, the Alzheimer’s Neuroimaging Initiative. Nature Communications Biology. doi: 10.1038/s42003-021-02133-x.